How does grt achieve global routing?

[Icamd]

Fastroute is a great global routing algorithm that can complete the routing of millions of two-pin nets in such a short time. However, after reading the Fastroute paper, it does not mention how the program allocates threads for the millions of two-pin nets. Does Fastroute divide all the two-pin nets into multiple portions and assign a thread to each portion? How are these threads synchronized to update the capacity of the whole global routing grid? Thank you!

[maliberty]

FR is not multi-threaded.

[Icamd]

If Fastroute is not multi-threaded, does it process millions of individual two-pin nets one by one? How does it achieve such incredible processing speed? Thank you

[maliberty]

yes serially. You can read the various papers at https://www.engineering.iastate.edu/~cnchu/FastRoute.html which describe the algorithms.